Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 357 A The Development of an Individual Potential for Violence This appendix does not purport to be a state-of-the-art review of individual- level influences on the development of violent behavior; that would require a book rather than a comparatively short paper. It aims to summarize briefly some of the most important findings and theories, but more especially to identify key issues and questions that are unresolved and to recommend ways of resolving them. The recent book edited by Pepler and Rubin (1991) provides more extensive information about findings and theories. The likelihood of someone's committing a violent act depends on many different factors. Biological, individual, family, peer, school, and community factors may influence the development of an individual potential for violence. Whether the potential becomes manifest as a violent act depends on the interaction between this violence potential and immediate situational factors, such as the consumption of alcohol and the presence of a victim. The focus in this appendix is on factors that influence the development of a person with a high potential for violence in different situations. The emphasis is especially on individual-level factors, such as temperament, IQ, and impulsivity, although other factors (family, peer, and school) that influence the development of a potential for violence are also discussed. Nonmanipulable individual factors such as sex and ethnic origin are not discussed, except insofar as they interact with other factors. It seems probable that the greater likelihood of males and
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 358 blacks to commit violent offenses might be explained by reference to some of the other factors discussed here. Most of the developmental research on aggression and violence has been carried out with males. Many facts about the development of violence are so well known and well replicated that they hardly need to be reviewed here. In particular, it is clear that aggressive children tend to become violent teenagers and violent adults. In other words, there is significant continuity over time between childhood aggression and adult violence. For example, Farrington (1991b) showed that aggressive children at ages 8-10 and 12-14 (rated by teachers) tended to have later convictions for violence and to be violent at age 32 (according to their self-reports). These kinds of results make it plausible to postulate that the ordering of people on some underlying construct of individual difference such as violence potential is tolerably consistent over time. However, it is also true that people change and that it is important to investigate how and why they change. These results also mean that knowledge gained about the correlates, predictors, and causes of childhood aggression is relevant to the explanation of teenage and adult violence. Many factors present early in life (such as childhood temperament or family influences) may predict adult violence essentially because they influence the development of childhood aggression and because childhood aggression tends to develop into adult violence. The predictors and correlates of childhood aggression and adult violence are so well known and well replicated that they are not reviewed in detail here. For example, Farrington (1989) showed that they include individual-level factors such as low IQ, low school attainment, high impulsivity, and poor concentration and family factors such as low family income, large family size, convicted parents, harsh discipline, poor supervision, and separations from parents. This appendix begins by discussing some important conceptual issues that need to be resolved, focusing on questions rather than answers. It then briefly summarizes key results, theories, and interventions before concluding with some recommended research priorities. Conceptual Issues Are there specific subtypes of violence that need to be explained by different theories? An important issue is whether violence should be treated as a homogeneous or heterogeneous category.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 359 Our preference is for heterogeneity, but in discussing development we usually refer to violence in general. The focus of the panel is on intentional physical assault. Within this context, the most important type of violence that has been studied in developmental research is assault (including wounding), and most of our knowledge is essentially about this phenomenon. Robbery has also been studied quite extensively. The distinction between assault and robbery mirrors that commonly drawn between angry or hostile (emotional) aggression and instrumental violence (e.g., Berkowitz, 1978). Although we focus on the development of a potential for violence rather than the violent act, it is important to recognize the existence of subtypes of violent acts, such as sexual and nonsexual violence. There has been less developmental research on homicide due to its rarity in community samples (for an example, see Solway et al., 1981). It is unclear whether homicide is qualitatively or quantitatively different from assault and wounding (see below). There has been a great deal of research in recent years on violence to spouses/cohabitants and children, but comparatively little of it has focused on the development of the offender. To what extent is violence merely one element of a more general syndrome of antisocial behavior? Information about violence is usually obtained from studies of offending in general. There are relatively few studies specifically on violence (e.g., Miller et al., 1982; Hamparian et al., 1978). Generally, violent offenders tend also to commit nonviolent crimes (see the section below on violent crimes in criminal careers). That is, offenders are versatile rather than specialized. Furthermore, there may be a general syndrome of âantisocial personality," which persists from childhood to adulthood and is characterized (in addition to psychopathic traits) by a wide variety of antisocial acts of which violence is one type (Farrington, 1991a). The psychiatric categories of childhood conduct disorder and adult antisocial personality disorder both contain indicators of aggression and violence (American Psychiatric Association, 1987). There has been quite a lot of developmental research on the categories of psychopathy and antisocial personality disorder (e.g., McCord and McCord, 1956; Robins, 1991). It is important to note that while psychopaths commit a disproportionate number of violent offenses, not all psychopaths are violent and, conversely, not all violent offenders are psychopaths (Hare, 1981). Features of psychopathy include lack of remorse and guilt, callousness and lack of empathy, egocentricity, impulsivity, early antisocial behavior,
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 360 and the commission of repeated and varied types of criminal acts (Hare, 1981). Psychopaths are unusually persistent offenders, continuing to offend at a high level until age 40 (Hare et al., 1988). Psychopathy has also been investigated extensively at the biological level, including genetic (Baker, 1986), hormonal (Virkkunen, 1987), psychophysiological (Hare, 1978), and neuropsychological (Raine and Venables, in press) factors. What are the manifestations of violence potential at different ages, and are they part of a developmental sequence? As already mentioned, there is continuity between indicators of childhood aggression and adult violence, suggesting persistence of an underlying construct that is here labeled violence potential. This construct may have different manifestations at different ages. It is important to attend to both homotypic continuity, in which violent behavior per se is preserved, and heterotypic continuity, in which early violence in the teenage years is replaced in adulthood with a different form of behavior that is not obviously violent. Such transformations are common in both psychological and biological development (for a useful discussion of concepts of developmental continuity, see Kagan, 1980). It is important to investigate sequences of onsets of different kinds of offenses and deviant (e.g., aggressive) behaviors, and to determine the probability of one behavior's following another and the average time interval between onsets. This information should be the starting point in trying to answer questions about why one behavior follows another. Conceptually there are three main reasons for developmental sequences. First, different acts may be different behavioral manifestations of the same underlying construct at different ages (e.g., an antisocial tendency manifests itself first in shoplifting, later in burglary, and later still in the abuse of spouses and children) but with no facilitating effect of an earlier act on a later one. Second, different acts may be different behavioral manifestations of the same or similar constructs at different ages and also part of a developmental sequence, in which one act is a stepping stone to or facilitates another act (e.g., smoking cigarettes tends to lead to marijuana useâKandel and Faust, 1975). Third, different acts may be indicators of different constructs and part of a causal sequence, in which changes in an indicator of one construct cause changes in an indicator of a different construct (e.g., school failure leads to truancy). The first of these ideas can be distinguished empirically from the second and third. If acts in a sequence are all different behavioral
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 361 manifestations of the same construct (like symptoms of an illness), then preventing or changing an early act in the sequence will not necessarily affect the probability of later acts unless there is some change in the underlying construct. However, with developmental and causal sequences, changing an early act in the sequence will affect the probability of later acts. It is harder to distinguish the second and third ideas empirically, as the key distinction between them is conceptual. Some key issues are as follows: How can an underlying theoretical construct such as violence potential best be operationally defined and measured at different ages? Does it have different manifestations at different ages, and if so what are they (e.g., cruelty to animals at age 6, assaults on classmates at age 12, wounding at age 19)? What are the most common developmental stepping-stone sequences that include adult violence? For example, does hyperactivity at age 4 lead to conduct disorder at age 8, and then to adolescent aggression and adult violence? How can we establish the existence of such sequences, and distinguish them from causal sequences and from different manifestations of the same underlying construct? To what extent is violence potential consistent over time? A key question is: What constructs underlie aggressive and violent behavior, and how general or specific should they be? Should we assume that all persons can be ordered on a dimension of violence potential at a given age, or that they can be ordered on a more general dimension such as antisocial personality or "potential for antisocial behavior"? Alternatively, violence could be viewed as a categorical variable, with violent people differing qualitatively rather than quantitatively from nonviolent individuals. It is important to investigate how violence and violence potential vary with age. Studies show that the most violent people at one age tend also to be the most violent at another. Hence (provided that this does not entirely reflect consistency in the environment), there seems to be consistency in the relative ordering of individuals on the underlying dimension of potential for violence, as already mentioned. However, absolute levels of violence (or violence potential) may vary considerably with age, for example, decreasing markedly after the teens and early twenties. It is important to explain both relative stability and absolute change (Farrington, 1990). Many projects show significant stability and continuity for aggressive behavior after ages 7-8. In all, 16 separate studies with lags ranging from 6 months to 21 years have reported stability
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 362 coefficients for aggression ranging from .36 to .95 (Olweus, 1979). Similarly, Huesmann et al. (1984) reported a stability coefficient of .46 between ratings of peer-nominated aggression at age 8 and criminal offenses at age 30. Furthermore, such stability in aggression appears to be consistent across cultures (Eron and Huesmann, 1987). Eron and Huesmann (1990) suggest that such strong stability is a product of continuity of both constitutional and environmental factors. The reasons for the strong stability need to be investigated in more detail. There is important continuity between juvenile and adult violence, as shown for example in the follow-up by Hamparian et al. (1985). In other words, violent behavior that begins before the eighteenth birthday tends to continue after it. Are extreme forms of violence (e.g., homicide) different in degree or in kind from other forms of violence? Generally, all types of violence tend to be interrelated in the sense that people who commit one type have a relatively high probability of also committing other types. Again, this is in conformity with the idea of an underlying violence potential. However, this is perhaps least true of homicide. Some homicides are committed by people with a history of violence, but others are committed by relatively nonviolent people under extreme stress or provocation (see the distinction between undercontrolled and overcontrolled individuals, e.g., Megargee, 1973). Hence, homicide in some cases seems an extreme point on a continuum of violence and in other cases seems qualitatively different from other (less serious) types of violence. This notion is consistent with the view that violence, like any complex behavior, has multiple and heterogeneous etiologies, and that there is no unitary type of person who is violent. Just as psychiatric categories such as depression and schizophrenia are heterogeneous rather than unitary concepts, violence also is likely to be heterogeneous, if not more so. If this is the case, then an important goal of future research should be to delineate subtypes of violence that may have different etiologies. We propose to focus more on the development of an individual potential for violence than on the occurrence of the violent act. Generally, violence arises out of an interaction between persons and situations. Some people are consistently more likely to be violent than others in many different situations (just as some situations or environments are consistently more likely than others to elicit violence from many different persons). While characteristic forms and amounts of aggression change dramatically with
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 363 age (Eron et al., 1983), the aggressiveness of an individual relative to the rest of the population may remain fairly stable. For example, a child who is at the top of the distribution for aggression at age 8 is likely to be near the top of the distribution 20 years later. The focus here is on the violent persons indicated by these cross-situational and cross-time consistencies. Are the results obtained by studying within-individual differences in violence over time different from those obtained by studying between-individual differences in violence? The prevalence and correlates of aggression and violence at different ages can be studied either within individuals (using longitudinal data) or between individuals (using cross-sectional data). Key issues are the following: How can we explain within-individual changes over age in violence and violence potential, as well as between-individual variations at different ages? How do biological, individual, family, peer, school, and community factors influence within-individual changes and between-individual variations in violence and violence potential? Existing criminological theories are not very useful because they typically focus only on between-individual variations in the teenage years, and they lack a concern with either development or violence. In any case, there is a need for basic developmental information that might inform future theories. Influences On Violence It should be noted at the outset that no one influence in isolation is likely to account for the development of a potential for violence, except perhaps in some special cases. It is possible, for example, that to produce a violent adult, one needs, at a minimum, a child born with a particular temperamental profile, living in a particular family constellation, in a disadvantaged neighborhood, exposed to models of aggression and patterns of reinforcement of aggressive behavior, having a particular school experience, having a particular set of peer relations, and also experiencing certain chance events that permit the actualization of violent behavior. Furthermore, it must be remembered that the influences listed below must be viewed within the context of other (biological and social) influences. In particular, factors such as sex hormones, endocrinological factors, psychophysiological factors, and drugs and alcohol might all be expected to interact in complex ways with individual influences. We know that future adult violence can be predicted, to a statistically significant degree, on the basis of childhood or adolescent
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 364 factors (Farrington, 1989). Generally, predictive analyses focus on the additive or interactive effects of factors, while causal analyses focus on their independent effects. However, there is a great need for more detailed research on how predictability varies with the ages at which the predictors and outcomes are measured. Specific childhood indicators of aggression include temper tantrums in infancy, physical aggression toward peers and siblings, cruelty to animals, disobedience, quarreling, and disruptiveness in class; prechildhood predictors include pregnancy and birth complications. Prenatal/Perinatal Factors Several studies have reported a relation between prenatal/perinatal factors and later violence. Litt (1971) found that perinatal trauma was predictive of impulsive criminal law offenses in a cohort of nearly 2,000 consecutive births in Denmark. Lewis et al. (1979) similarly found that aggressive, incarcerated delinquents were more likely to have sustained perinatal trauma than less aggressive, nonincarcerated delinquents. Mungas (1983) reported a significant relation between violence and perinatal factors within a group of psychiatric patients. While these studies have indicated a main effect of perinatal factors, one study suggests that prenatal/perinatal factors may interact with social factors in increasing the risk for later violence. Mednick and Kandel (1988) found that minor physical anomalies (an indicator of first-trimester pregnancy complications) were predictive of the number of violent offenses at age 21, but only in children raised in unstable, nonintact homes. Minor physical anomalies have also been found to be elevated in hyperactivity, which itself is related to later delinquency and aggression (Fogel et al., 1985). Similarly, perinatal difficulties have been found to be predictive of aggressiveness (bullying and fighting) at age 18 but, again, this was true only for those raised in unstable home environments. While these data suggest that a link may exist between pregnancy/birth complications and later violence, there has been almost no research on the pathways by which prenatal and perinatal disturbances might lead to violence. It is possible that obstetric complications may result in damage to brain mechanisms that act to inhibit violent behavior (e.g., the prefrontal cortex). Alternatively, violence may be a by-product of impulsivity, hyperactivity, or cognitive deficits that are themselves produced by damage
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 365 to the central nervous system resulting from these complications. Possible interactions between pregnancy/birth complications and the early rearing environment are of potential importance, since they suggest that a stable early home environment may protect a child from the negative effects of these complications. For example, it has been observed that being raised in a stable home protects a child from the cognitive deficits otherwise associated with prematurity (Drillien, 1964). These data are at present limited, and confirmation and extension of preliminary studies are required in order to uncover the possible mechanisms by which pregnancy and birth complications may lead to the development of later violence, as well as to establish further the interaction of these complications with social and cognitive factors. In particular, experimental studies that provide greater prenatal care to mothers at risk for pregnancy and birth complications would help to test whether there is a causal connection between such complications and later violence. An interesting feature of these preliminary findings is that the effect of pregnancy and birth complications is relatively specific to later violence. Such complications do not, for example, appear to be predictive of nonviolent criminal behavior (Mednick and Kandel, 1988). This is of some importance since (as noted earlier) there is relatively little research on factors that selectively predict violence as opposed to nonviolent crime or crime in general. Temperament Early temperament may well constitute a risk factor for later aggressive and violent behavior. When confronted with unfamiliar situations (e.g., encounters with unfamiliar children and adults), some children ages 20-30 months tend to be shy, vigilant, and restrained ("inhibited"), whereas others tend to be sociable, spontaneous, and relatively fearless in their behavior ("uninhibited") (Kagan, 1989). Approximately 30 percent of children of this age can be classified into either inhibited or uninhibited categories. These temperamental differences at 21 months are relatively stable, with 75 percent of children being similarly classified as inhibited or uninhibited at 7.5 years. In addition, children exhibiting the most extreme forms of behavior at age 21 months are the most likely to be similarly classified at age 7.5 years. These two temperamental "types" have been found to differ in physiological terms. Inhibited children have higher and more stable heart rates than uninhibited children (Reznick et al., 1986) and show greater heart
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 366 rate acceleration during cognitive testing. A global physiological index of arousal made up of eight measures including heart rate, pupil dilation, norepinephrine, cortisol, and vocal tension correlated significantly with inhibition at age 21 months (r = .70) and at age 7.5 years (r = .64), with the uninhibited children being more underaroused. The significance of this research for violence lies in the possibility that a fearless, uninhibited, early temperament may be a risk factor for later aggression and violence, especially in children with low socioeconomic status, whereas fearfulness may act as a protective factor against aggression. Temperament may explain why only a proportion of children from high-risk homes and neighborhoods develop antisocial or violent behavior (Kagan, 1991). The fact that fearless young children have been found to have low heart rates is consistent with findings of low heart rates in undersocialized children ages 7-15 with aggressive conduct disorder (Raine and Jones, 1987), with the demonstration that persons convicted of violence tend to have low heart rates (Farrington, 1987; Wadsworth, 1976), and with the fact that low heart rate is thought to reflect a factor of fearlessness (Raine and Jones, 1987). Prospective longitudinal research and cross-sectional studies measuring early temperament in conjunction with measures of physiological arousal and aggression/violence across the life span are required to establish how far early temperament and underarousal are predisposing factors for later aggression and violence. The ability of measures of temperament and heart rate taken as early as 20 months to relate to later violence in adolescence and adulthood is clearly an important topic to be addressed. Although such factors in isolation may not be expected to be strong predictors of violence, in conjunction with other early family and cognitive measures, the degree of prediction may be considerable. The speculation by Kagan (1989) that physiological differences between shy and fearless children may reflect differential thresholds of limbic structures such as the amygdala and hypothalamus is in agreement with research implicating limbic structures in aggression in both animals and humans. Testing in violent and nonviolent subjects using positron emission tomography (PET) techniques could yield important results. Individual Factors Violent offenders tend to have certain personality features as children. In particular, they are high on hyperactivity-impulsivity-attention
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 367 deficit, tend to be restless and lacking concentration, take risks, show a poor ability to defer gratification, and have low empathy (e.g., Farrington, 1989). They also tend to have particularly low IQ scores (more so than other types of offenders). Physical child abusers have been found to have low self-esteem, a negative self-concept, feelings of inadequacy and inferiority, and an external locus of control (Milner, 1986, 1990). They also have psychopathological personality characteristics, higher trait anxiety and trait anger scores, higher neuroticism scores, and elevations on psychopathic, mania, paranoia, and schizophrenia scales of the Minnesota Multiphasic Personality Index (Milner, 1986). One longitudinal study indicated that although IQ at age 8 predicts aggression at age 30, IQ is no longer a significant predictor after the effects of aggression at age 8 are removed (Huesmann and Eron, 1984). It is possible that low IQ at an early age contributes to the early adoption of aggressive behavior, and that once such behavior is firmly established its further development into adult violence is relatively unaffected by low intellectual functioning (Huesmann and Eron, 1984). Family As already mentioned, violent offenders tend to have experienced poor parental childrearing methods, poor supervision, and separations from their parents when they were children (Farrington, 1991b). In addition, they tend disproportionally to come from low-income, large-sized families in poor housing in deprived, inner-city, high-crime areas. (This appendix does not discuss community influences on crime.) Recent research has focused on the link between being a victim of physical abuse and neglect as a child and later violent offending. Although it is commonly assumed that individuals who experience physical abuse as children grow up to become violent adults, there are surprisingly few sound empirical data to support this assumption. Widom (1989a), in a critical review of seven different areas of research that bear on this question concluded that most studies were methodologically weak. Her main criticisms include overreliance on self-report and retrospective data, inadequate documentation of child abuse, weak sampling techniques, and infrequent use of control groups. The soundest study conducted to date on physical abuse of children employed a large matched (case control) cohort design to
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 368 study the effects of different forms of early abuse (defined by court reports) on later crime and violence (Widom, 1989b). In comparison to controls, abused and neglected children had a significantly greater number of arrests for violence as adults. These effects held for males but not for females, and for blacks but not for whites. Neglect in childhood without violence was just as likely to lead to an arrest for violence in adulthood as was physical abuse in childhood. Although these findings are consistent with a causal relation between early abuse and later violence, further research is required to rule out other possibilities. It is possible, for example, that the experience of early abuse does not itself cause adult violence. It may be that parental abuse is a response to the child's aggressive behavior, since Eron et al. (1991) found that the best predictor of adult aggression was the extent of childhood aggression, regardless of parental behavior. Furthermore, parents who physically abuse their children may pass on to their children a genetic or biological predisposition to violence that is responsible for the later violence seen in these children (DiLalla and Gottesman, 1991; Widom, 1991). An important issue is whether causal effects are immediate or delayed. For example, we know that abused children tend significantly to become violent adults. Is this an example of a long-delayed "sleeper effect," or is it mediated by a chain of shorter-term causes and effects? If so, what are they? As already mentioned, numerous studies show that violent offenders tend to come from certain types of family backgrounds. In particular, they tend to have been subjected to physical punishment, they tend to have alcoholic or criminal parents, and they tend to have disharmonious parents who are likely to separate or divorce. Physical child abusers also tend to have parents who lack warmth, are rejecting and hostile, and use reinforcement in unpredictable ways (Milner, 1990) and are less likely to report having had a caring adult/friend in their childhood (Milner et al., 1990). These features of parents may act as mediators between physical abuse in childhood and adult criminal violence. Each year, 30 percent of female homicide victims are killed by their husbands or boyfriends (Federal Bureau of Investigation, 1982), and over a million abused women seek medical help for their injuries (Stark and Flitcraft, 1982). Most of the research on the etiology of spouse abuse has focused on intrapersonal as opposed to interpersonal or sociocultural factors. This research suggests that wife batterers have witnessed abuse in their own families of origin, have low self-esteem, tend to abuse alcohol, and have low
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 369 assertiveness (Margolin et al., 1988). These factors, however, also characterize nonviolent criminals, and research is needed to understand the factors that may possibly link experience with early child abuse with later wife abuse. Unfortunately, research into the etiology of wife battering has involved women from small, nonrepresentative samples residing temporarily in shelters. Margolin et al. (1988) have recommended that future research in this area should (a) be directed at nonclinical and noncriminal samples, (b) focus on the development of typologies with a view towards intervention, (c) be developed to test theories of wife battering, and (d) aim to assess what factors predict such violence in a longitudinal context. Beginning around age 5 or 6, children identify with their parents and with ethnic and class groups. By identification we refer to children's beliefs that they share the psychological properties of the persons or groups with whom they are identified. This universal process is an important part of psychoanalytic theory. It was also recognized by commentators on human nature writing long before Freud was born. Because of identification, children who live with parents who possess undesirable qualities (unemployed, unfair, unjust) will believe that some of these properties also belong to them. As a result, they will establish a conception of themselves as undesirable or as incapable of obtaining desired goals. Such a state is likely to create psychological conditions for later aggressive behavior. One reason why race may be associated with violence in our culture is that black children who live in poverty may identify both with the undesirable qualities of being poor as well as with their ethnic group, which they perceive to be the target of hostility by the white majority. These identifications may be part of the explanation why black children are disproportionally at risk for aggressive behavior. Peer and School Factors Coie et al. (1991) has argued that aggressiveness was the single most important reason for a child to be rejected by peers, pointing to the fact that 30-40 percent of socially rejected children were highly aggressive. It is uncertain, however, to what extent peer influences are important in the development of violent offending, and whether rejection by peers causes aggressive behavior, whether aggression causes social rejection, or both. There is some evidence in favor of aggression's causing rejection, which supports Coie's argument (Huesmann and Eron, 1986). However, if rejection
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 370 causes aggression, then interventions aimed at improving children's social relations with their peers might lead to reductions in aggressive behavior. Olweus (1991) has estimated that 9 percent of schoolchildren in Norway and Sweden are regular victims of bullying, while 7-8 percent of children engage in bullying. It is believed that four factors contribute to the development of bullying behavior in school: (a) lack of warmth and involvement by the primary caretaker, (b) permissiveness regarding aggressive behavior by the caretaker, (c) the use of physical punishments and violent emotional outbursts as childrearing methods, and (d) an âactive and hot-headed" temperamental predisposition in the child (Olweus, 1980). It might be expected that school bullying would predict adult violence, but there appears to be little hard evidence for this at the present time. However, the finding that children who at age 8 are rated by their peers as being aggressive in school have significantly more indications of antisocial aggression as adults than other children would support this expectation (Eron and Huesmann, 1990). A statistical survey of school crime between 1975 and 1977 indicated that 156 out of 1,000 secondary school students reported having been assaulted. While 42 percent of the student assaults and 11 percent of the student robberies resulted in injury, most injuries were relatively minor (U.S. Department of Health, Education, and Welfare, 1978), and in the past weapons have been infrequently used in such assaults and robberies (McDermott and Hindelang, 1981). Only 1 in 20 student and teacher victimizations resulted in an injury (McDermott and Hindelang, 1981). Spatial characteristics of schools can influence violence in that (a) relatively high numbers of individuals occupy a limited amount of space, (b) the capacity to avoid confrontations is somewhat reduced, (c) the imposition of behavioral routines and conformity may contribute to feelings of anger, resentment, and rejection, and (d) poor design features may facilitate the commission of violent acts. These features of schools suggest possible interventions for reducing school violence (e.g., enhancement of environmental design, promoting student participation in rule enforcement), but to date such suggestions have been largely impressionistic and anecdotal. Violent offenders tend to have a history of school failure, including low intelligence, poor attainment, and truancy (Farrington, 1991b).
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 371 Effects of Television Violence Most reviews of the literature (e.g., Heath et al., 1989) conclude that exposure to television violence encourages aggressive behavior by children and adults. Even reviews restricted to naturalistic field experiments on aggression, which should have the highest internal and external validity, reach the same conclusion (e.g., Wood et al., 1991). However, there are some differences in interpretation and numerous methodological problems, for example those brought out in the debate between Friedrich-Cofer and Huston (1986) and Freedman (1986). The effects of television violence on aggression may be short term (e.g., activation, arousal, disinhibition, contagion) or long term (e.g., changes in behavior, attitudes, or values), but most research focuses on short-term effects. An extensive meta-analysis prepared for the panel by Comstock and Paik (1990) covered 188 studies and 1,126 comparisons between 1957 and 1990. They found that the average effect size (in this case, a correlation) was .31. This effect is quite substantial. For example, in a 2 by 2 table, it might correspond to 51 percent of those exposed to television violence showing aggressive behavior, in comparison with only 20 percent of a control (nonexposed) group (Farrington and Loeber, 1989). Laboratory experiments (.40) showed higher average effect sizes than field experiments (.27) or surveys (.19), and effect sizes were greater for physical violence (.23) and robbery (.28) than for criminal violence (.10). Overall, the vast majority of studies, whatever their methodology, showed that exposure to television violence resulted in increased aggressive behavior, both contemporaneously and over time. Life Transitions It is important to investigate whether key life events or transitions influence within-individual changes in violence or violence potential over time, such as puberty, moving home or changing schools, graduating from school or dropping out, starting a job or becoming unemployed, getting married, having a child, and so on. Most of these influences have been studied with respect to offending in general rather than specifically to violent offending. The onset and velocity of puberty vary considerably in females and in males, although it is more common to consider these changes in the course of female than male development because of their more dramatic bodily changes. The variation is such that a substantial number of children can be classified as either early or late maturers. In measuring the behavior of young adolescents it is
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 372 essential to note that the biological changes accompanying puberty are confounded with psychological changes and adjustments in schooling, given the customary switch from elementary to junior high school that occurs around age 13. Early maturation is generally thought to be a disadvantage in girls (e.g., they may be attracted to older males, older males find them attractive), while it appears to be an advantage in boys (e.g., the increase in muscle mass and coordination contributes to athletic skill). Magnusson et al. (1986) found that girls with an earlier age of menarche committed more antisocial acts than other girls, as a function of having older friends. During and after puberty an association between increased levels of testosterone and aggressive behavior in males has been demonstrated (Olweus, 1987). Much aggressive behavior typically has its origin in the period prior to the onset of puberty, making it unlikely that rising hormone levels at puberty have a causal influence on onset. Hormone levels are more likely to be related to the frequency or intensity of aggression. It is not clear how variations in gonadal hormones are affected by changes in the social environment, nor is the interaction between gonadal and adrenal sex hormones well understood. In fostering a developmental approach to the study of early determinants of aggressive and violent behavior, these areas are ripe for investigation. Protective Factors It is important to investigate how protective factors interact with other influences at the individual level that encourage the development of violence, as well as with influences at the biological and social levels. There are two main definitions of protective factors (see, e.g., Farrington et al., 1988). A protective factor may be merely the opposite end of the continuum to a risk factor. For example, just as low IQ is regarded as a risk factor, high IQ can be regarded as a protective factor. Alternatively, a protective factor may be a variable that interacts with a risk factor to minimize its effect. For example, if low IQ predicted violence among low-income families but not among high-income families, high income might be protecting children from the effects of low IQ. There have been several studies focusing on why children at risk of offending do not become offenders, and on what the protective factors are. Rutter and Giller (1983) reported that certain life events (e.g., changes in peer group, leaving school, moving away from home, marriage) are associated with reductions in delinquency
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 373 and crime. Research by Farrington et al. (1988) indicates that the main protective factors seemed to be shyness, nervousness, or social isolation. In other words, the vulnerable boys from typically criminogenic backgrounds who did not become offenders tended to be socially withdrawn and isolated in childhood (and indeed later in adulthood as well). Such findings are consistent with the possibility that a shy, fearful temperament in early life may represent a protective factor for later aggression (Kagan, 1989). However, research by McCord (1987) suggests that childhood aggressiveness coupled with shyness was more likely to be followed by crime than aggressiveness without shyness, and similar results were obtained by Kellam et al. (1983). Huesmann and Eron (1988) found that aggressive children at age 8 who did not commit later crimes by age 30 differed from the aggressive 8-year- olds who accumulated criminal records in having a higher IQ and in having parents who regularly attended religious services; these were the only two group discriminators after controlling for occupation, income, family size, and geographic residence (Huesmann and Eron, 1988). Other identified protective factors include being firstborn and coming from smaller families characterized by low discord (Werner and Smith, 1982). Violent Crimes In Criminal Careers It is important to investigate how far the "criminal career perspective" (Blumstein et al., 1986) might be useful in explaining the development of criminal violence. This perspective focuses on such concepts as onset, continuation, and desistance; specialization and versatility; escalation in seriousness; and frequency of offending by active offenders. It is important to determine the relationship among these concepts. For example, do those who commit criminal violence at a relatively early age tend to commit violent offenses at high rates and tend to have long careers of violence? Most studies of violence are based on official records of arrests or convictions, which suggest that violence is a relatively uncommon event. However, self-reports of offending show that physical assault is quite prevalent. For example, Farrington (1989) reported that only 12 percent of the inner-city males he studied were convicted of violent offenses between ages 10 and 32, but as many as 37 percent were involved in physical assaults between ages 27 and 32, when violence was decreasing.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 374 Very few projects permit the investigation of the occurrence of violent crimes in criminal careers. Generally, self-report studies with general population samples (e.g., Elliott et al., 1989) are unsuitable, for two main reasons. First, the prevalence of unambiguously violent crimes (as opposed to "gang fights") is very low, possibly because many of the most violent people in the population are not included in the samples interviewed. Second, information about the time ordering of different types of offenses is not collected in such studies. It would be desirable in future self-report studies of offending to collect time ordering information about unambiguously serious offenses, which are rare; it would be impractical to collect such data about all self-reported offenses. Self-report or official record studies with retrospectively defined offender samples (e.g., Miller et al., 1982) are also unsuitable, because of the difficulty of reconstructing the criminal careers of any given birth cohort. This leaves prospective cohort studies with official record information about offending as the most suitable for studying violent crimes in criminal careers, but unfortunately many of these studies (e.g., Wolfgang et al., 1972) stop at the eighteenth birthday, before the peak of violent offending in the young adult years (18-25). All results, of course, are greatly affected by the age of truncation of the study, and there is the continuing problem in U.S. studies of linking juvenile and adult records. Also, the sample sizes in many studies are inadequate to investigate violent offending, because of its low prevalence (especially outside the United States). In his detailed review of knowledge about violent criminal careers, Weiner (1989: Table 2.1) listed 44 relevant studies. However, only three American data sets could provide minimally adequate information about violent crimes in criminal careers: (1) the 506 Cambridge-Somerville (Boston) males born about 1928 and followed up to age 45 by McCord (1980), (2) the 975 Philadelphia males born in 1945 and followed up to age 30 by Wolfgang et al. (1987), and (3) the 633 continuously resident Racine children born in 1942 and followed up to age 32 by Shannon (1988). In addition, four foreign data sets could provide minimally adequate data: (1) the 411 London males born mostly in 1953 and followed up to age 32 by Farrington (1991b), (2) the 710 Swedish males born about 1955 and followed up to age 30 by Stattin et al. (1989), (3) the 15,117 Stockholm children born in 1953 and followed up to age 25-26 by Wikstrom (1985), and (4) the 28,879 Copenhagen males born in 1944-1947 and followed up to age 27-30 by Moffitt et al. (1989). The lack of adequate data about violent crimes in
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 375 criminal careers makes it easier to raise questions than to answer them. The main violent crimes that have been studied are homicide, forcible rape, robbery, and assault. It would be desirable to study different types of violent crimes in criminal careers, but this is very difficult. For example, it would be desirable to know how different types of violent crimes are interrelated and whether, for example, homicide offenders have a significant tendency also to commit forcible rape, robbery, or assault. The most suitable data set for answering this question is that collected by Miller et al. (1982), but they do not seem to have addressed it. It would be desirable to know whether these violent crimes were typically committed in a certain order or at certain ages, but available data limit this review to violent crimes in general rather than different types of violent crimes. Some key questions follow: (1) What proportion of crimes in criminal careers are violent? Is it safe to conclude that only a small proportion are violent? For example, only 12 percent of offenses leading to conviction in the London cohort (85 of 683) were violent, as were only 5 percent of offenses recorded by the police in the Stockholm cohort (1,290 out of 23,774). The 633 Racine children yielded only 21 assaults and 7 robberies, so that only 1 percent of their offenses were violent felonies; only 9 percent of the Philadelphia crimes involved injury. (2) What proportion of offenders include a violent crime in their criminal careers? These percentages, of course, are higher than those in question 1, but it is still safe to conclude that only a minority of offenders include a violent offense in their criminal careers. For example, only 33 percent of convicted males in the London cohort (50 of 153) were convicted of violence, representing 12 percent of all cohort males. In the Stockholm cohort, 21 percent of recorded offenders were recorded for violence (591 of 2,837), representing 4 percent of the cohort; and in the Copenhagen cohort, 7 percent of recorded offenders were recorded for violence (735 of 10,918), representing 3 percent of the cohort. Even in the Cambridge- Somerville study, only 14 percent of the cohort (69 of 506), or 18 percent of the offenders, were convicted of violent crimes up to age 45. (3) How many violent crimes are committed by violent offenders? Studies outside the United States show that the majority of officially recorded violent offenders are recorded for only one violent offense. This was true, for example, of 72 percent of the Stockholm violent offenders, 76 percent of the Copenhagen violent
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 376 offenders, and 70 percent of the London violent offenders. However, in the Miller et al. (1982) study in Ohio, the violent arrestees included 20 percent arrested 5 or more times for violence and 53 percent arrested 2-4 times for violence, so the same conclusion may not hold in the United States. (4) How many nonviolent crimes are committed by violent offenders? Is it safe to conclude that violent offenders commit more nonviolent crimes than violent crimes? In the London cohort, the 50 violent offenders were convicted of 85 violent crimes (average 1.7) and 263 nonviolent crimes (average 5.3), so that only 24 percent of their crimes were violent. Nearly all of these violent offenders (43 of 50) were also convicted for nonviolent crimes. Similarly, 86 percent of the recorded violent offenders in the Stockholm cohort (510 of 591) were also recorded for nonviolent crimes. In the Miller et al. study in Ohio, the violent offenders committed twice as many nonviolent crimes as violent crimes (8,368 as opposed to 4,163). (5) Are violent offenders merely frequent offenders? In other words, does the probability of committing a violent crime increase in direct proportion to the number of crimes committed? In the London cohort, this seemed to be true. Since 12.4 percent of all offenses were violent, it might be expected that 12.4 percent of the 49 cohort males who were convicted of only one offense would be convicted for violence, and the actual figure of 6 was close to the chance expectation of 6.1. Similarly, it might be expected that 23.3 percent of the 30 males who committed two offenses would commit at least one violent offense (1 - [.876]2), and the actual figure of 8 was close to the chance expectation of 7.0. Overall, the actual proportion of all offenders who were violent (33%) was not significantly different from the chance expectation (36%) assuming that violent and nonviolent offenses were committed at random. If there had been any specialization in violent offending, there would have been significantly fewer violent offenders than chance expectation, and they would have each committed more violent offenses on average than expected. These and other results led Farrington (1991b) to conclude that violent offenders were essentially frequent offenders. Similar conclusions were drawn by Miller et al. (1982) in Ohio and by Wikstrom (1985) in Stockholm. (6) Is there specialization in violent offending? While there is a high degree of generality in violent offending, there also appears to be some degree of specialization. For example, controlling for the number of offenses committed before and after the first violent
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 377 or property offense, Moffitt et al. (1989) in the Copenhagen cohort concluded that a first-time violent offender was 1.9 times as likely to commit a violent act among his future offenses as a first-time property offender. It is likely that the degree of specialization in violent offending is greater at adult ages than at juvenile ages. (7) What is the average age of commission of violent as opposed to nonviolent offenses? Generally, violent crimes are committed at a later age than most types of nonviolent crimes. In the London longitudinal study, violent crimes had the second-highest median age of commission (20) after fraud: higher than the more common burglaries or thefts (17). Violent crimes had a significantly higher age of commission than nonviolent ones, when comparisons were restricted to the 43 men with both violent and nonviolent offenses (31 committed violent crimes at higher ages, 12 committed nonviolent ones at higher ages). (8) What is the average age of onset of violent as opposed to nonviolent offenses? In the London study, violent crimes had a higher average age of onset than nonviolent ones, when comparisons were restricted to the 43 men with both violent and nonviolent offenses (35 violent higher, 6 nonviolent higher, 2 equal). It was the case for only 7 of the 153 convicted men that their first offense was a violent crime, making it impossible to determine whether the percentage of onsets that were violent changed with age. (9) What is the average serial number of violent as opposed to nonviolent offenses in criminal careers? In the London study, the average serial number of violent offenses was higher than for nonviolent offenses, when comparisons were restricted to the 43 men with both types of offenses (27 violent higher, 11 nonviolent higher, 5 equal). In other words, violent offenses tended to be committed relatively late in criminal careers. (10) What is the relation between juvenile violent offending and adult violent offending? Do the majority of juvenile violent offenders go on to commit violent offenses as adults and, conversely, were the majority of adult violent offenders previously convicted of violence as juveniles? In the London study, the majority of juvenile violent offenders did go on to commit adult violent offenses (7 of 10, with the other 3 committing adult nonviolent offenses). However, only a minority of adult violent offenders (7 of 47) were juvenile violent offenders. Of the remaining 40, 20 had juvenile nonviolent offenses and 20 had no juvenile offenses.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 378 (11) How useful are criminal career concepts such as career length and individual offending frequency as applied to violence? What is the probability of committing an (n + 1)th violent offense after committing an nth violent offense, and what is the average street time until the next violent offense? In the majority of studies in which the majority of violent offenders commit only one violent offense, ideas about careers of violence are not likely to be very useful. They could be useful in a study such as Miller et al. (1982), although this was not a prospective cohort study. In most projects, the main thrust of the research has to be to investigate the occurrence of odd violent offenses during a predominantly nonviolent criminal career. For example, it would be useful to investigate if violent offenses tended to follow certain types or sequences of nonviolent offenses. However, attempts to identify "violent career criminals" in most studies are doomed to failure. Theories This section briefly summarizes several theories that are particularly important to understanding of the development of violence and aggression. Several of these theories are complementary (e.g., social learning theory and cognitive-behavioral theory) and should not be viewed as mutually exclusive. Although the focus in this appendix is on developmental theories, it would also be interesting to investigate to what extent other theories (e.g., the rational choice theory of Clarke and Cornish, 1985) explain violent behavior. Frustration-Aggression Theory Frustration-aggression theory represents the first comprehensive theory of aggression that assigned a prominent role to learning (Eron, 1990). The basic principle underlying frustration-aggression theory is that, when people become frustrated (e.g., when their goals are thwarted), they become aggressive (Dollard et al., 1939). This theory fostered a great deal of empirical research testing hypotheses generated by it. One basic premise of the model, that frustration is a necessary facilitator of aggression, has been seriously questioned by a number of researchers (Eron, 1990). Other demonstrated weaknesses in this theory led to the development of research that placed a greater emphasis on external environmental cues as elicitors of aggression than on motivational or drive factors (Bandura, 1973).
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 379 Social Learning Theory Bandura (1973) proposed that aggressive behavior is learned and maintained through environmental experiences. He argued that aggression can be learned vicariously by watching (modeling), and that it can be inhibited if punishment succeeds in building up an association between aggression and anxiety (in which the anxiety might be interpreted as "guilt" or "conscience"). According to this theory, the maintenance of aggressive behavior is usually subject to the principles of reinforcement, so that behaviors that are reinforced will be repeated. Although research testing this social learning theory of aggression has generated mixed results, it has received some degree of empirical support (e.g., Eron, 1987). Cognitive-Behavioral Theories A cognitive perspective on aggression holds that aggressive behavior is a product of angry, aggressive thoughts. In this framework, interventions aimed at changing the occurrence of such thoughts would be expected to alleviate aggressive behavior. Huesmann and Eron (1989) have developed a cognitive model of the development of aggression that argues that such behavior is to a great extent controlled by programs or "scripts" learned during the child's early development. Such scripts suggest what events are about to occur, how the person should react to these events, and what the outcome will be. These scripts are retrieved from memory in response to appropriate environmental cues and are used to guide behavior. Under this model, a repeatedly aggressive child is one who consistently retrieves and employs aggressive scripts. The child's cognitive processes are also thought to be influenced by the parent's own cognitive processes, in that parents who view the world as hostile may reinforce their child's view of the world as hostile. Huesmann and Eron (1989) argue that aggressive behavior is stable because the scripts for aggression are themselves stable, due to the process of repeated rehearsal, whether through fantasizing, observation, or actual behavior. Dodge (1986) has developed an influential social information processing model of aggression. This model lays out a sequence of five cognitive operations involved in the development of aggressive behavior: (1) encoding, (2) interpretation, (3) response search, (4) response decision, and (5) enactment. Aggressive children appear to use fewer environmental cues to mediate behavior (Dodge and Newman, 1981) and tend to interpret the behavior of a peer as more hostile (Dodge, 1980). They are also less capable of
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 380 generating potential responses to conflict situations (Richard and Dodge, 1982) and are more likely to select passive and aggressive responses (Dodge, 1986). Interventions aimed at increasing the aggressive child's social information processing capabilities would, according to this model, have a significant effect in reducing levels of aggression. Dodge (1991) has recently drawn a distinction between reactive and proactive aggression, which is similar to Berkowitz's earlier distinction (1983) between emotional and instrumental aggression. This suggests that, while some aggressive children are troubling to others and use aggression in a proactive way to meet their goals (proactive aggression), others react in an angry, volatile manner and are troubled by others (reactive aggression). This distinction is supported by similar distinctions made by ethologists and psychobiologists between âaffective aggression," characterized by a high degree of autonomic arousal, and "hot-blooded" frenzied anger and between "instrumental aggression" characterized by low autonomic arousal and a cold-blooded, reward-seeking form of aggression. This two-factor model has received some support from factor- analytic studies (Dodge, 1991). While reactive aggressives at a young age are disliked by their peers, proactive aggressives are not necessarily disliked (Dodge and Coie, 1987). Proactive aggressives are also viewed as having leadership qualities and a better sense of humor than reactive aggressives (Dodge and Coie, 1987). Dodge (1991) speculated that these two forms of aggression have different neural substrates, have deficits in different stages of social information processing, and have different etiologies and developmental courses. Research Priorities Key Questions Important questions about the development of an individual potential for violence should be addressed: (1) Do male and black persons have a higher potential for violence than others and, if so, why? (2) To what extent do potentially violent people tend to seek out violent situations? (3) What are the developmental sequences linking violent offenses with other types of offenses and with childhood aggression? (4) How do individual, family, and school factors interact to
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 381 produce aggressive children and violent adults? (5) What are the factors that protect aggressive children from becoming violent adults, and what factors facilitate desistance from violent offending? (6) What are the effects of juvenile and criminal justice sanctions on violent offending? (7) Are there subtypes of aggression and violence (e.g., reactive and proactive aggression) that have distinct etiologies and are responsive to different forms of intervention? (8) Through what routes (e.g., temperament, cognitive deficits, social skills deficits) may pregnancy and birth complications lead to later violence, and how are these obstetric factors mediated by family and other environmental factors? (9) What factors contribute to the manifestation of fearful versus fearless early temperaments, and do early temperamental factors predispose children to later violent adult behavior? (10) What is the role of physical discipline in contributing to the development of violence? (11) Can reliable and valid indices of violence be developed? (12) What differences are there between people who commit violent acts and those who commit more general delinquent, criminal, or antisocial acts? Research Methods A variety of different research designs are useful in tackling these questions. For example, in investigating cross-situational consistency in violent behavior, short-term intensive longitudinal studies, with frequent data collection, would be useful. In studying factors that might inhibit aggressive children from becoming violent adults, intervention techniques such as skills training and sanctions should be evaluated in randomized experiments including follow-up periods. Prospective longitudinal studies would be needed to investigate developmental sequences from childhood aggression to adult violence. However, for rare acts such as homicide, retrospective or case control studies would be needed. Violence is a complex and multifaceted form of behavior. It is likely that there are no simple explanations of violence. It is essential that longitudinal research into violence should include measures of a wide range of variables, including biological, behavioral, social, and cognitive factors. Technological innovations are permitting deeper probing into questions of human biology, and the possibilities of breakthroughs in those areas are ever-present.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 382 Such research has implications for the interaction between influences at the individual level and biological influences. Longitudinal research does not necessarily have to extend from birth to adulthood. An "accelerated longitudinal design" could be used, in which a few overlapping age cohorts are followed up for a few years each (Tonry et al., 1991). Many correlates and predictors of violence have already been identified in past research. However, the precise ways in which they are linked to violence (e.g., through developmental or causal sequences) is not clear. In future research, frequent data collection is needed to track within-individual developmental sequences over time. Data from several different sources are needed (e.g., biological, individual, family, peers, school, community) to study the major causal influences on development and their interactive and independent effects. Special efforts should be made to investigate how the individual potential interacts with the environment to produce violent events. Very few studies begin with a sample of aggressive children and investigate possible protective factors influencing why some of them do not become violent adults. Such studies, however, would be capable of providing important insights regarding factors involved in desistance from violence. One valuable strategy for the future would be to combine longitudinal research studies of violence with experimental interventions, since such studies are capable of providing information on both the natural history and the course of development of violence together with causal effects of intervention methods (Farrington, 1988). A study by Huesmann et al. (1983) included an intervention with a subsample of subjects who were part of a larger longitudinal research design. The intervention was successful in reducing the aggression of the children receiving the treatment over a three-year period. In this continuing investigation the subjects who participated in the intervention at ages 8 to 10 will be reexamined to determine if the reduction in aggression has persisted over 15 years. In order to establish the effects of interventions on violence, subjects would ideally be assigned at random to intervention programs. It would also be desirable to conduct such studies in large cities, where the problem of violence is most acute. A good official record system, a competent research team, a large sample size, and methods to minimize attrition would be important prerequisites for such a study. Research on the success of interventions not only has important practical and policy implications but also provides important clues about causal effects, and it is to this research that we now turn.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 383 Interventions Prenatal, Perinatal, and Postnatal Health Care The link between prenatal/perinatal complications and later violence suggests that interventions to reduce violence could be targeted on the prenatal/ perinatal period. However, there appear to be no published studies reporting the effects of such early interventions on later violence. One possible intervention would be to provide prenatal and postnatal health care to an experimental group of mothers (e.g., as in the Infant Health and Development Program directed by J. Brooks-Gunn) and investigate its long-term effectiveness in reducing violence in comparison with a control group. Mothers in the experimental group could be provided with intensive antenatal care, including more frequent home visits by health professionals; education on diet, smoking, and alcohol during pregnancy; and earlier hospitalization to facilitate trouble-free delivery of the baby. Social support provided by midwives during pregnancy to socially disadvantaged mothers with a history of low-birthweight babies has been found to be effective in raising birthweight (Oakley et al., 1990). Postnatal interventions could be targeted on those within the experimental group who experienced perinatal complications in spite of the intervention. Low-birthweight and premature babies are more likely to experience speech and language disorders, neuromotor abnormalities (e.g., balance and coordination problems), perceptual problems, and visuo-spatial problems. These problems have themselves been implicated in the development of crime and violence (e.g., Moffitt, 1990). Early assessment for such difficulties in the target group coupled with interventions aimed to treat these conditions at an early age could be implemented. Low-birthweight babies are also more likely to suffer intellectual deficits than normal-birthweight babies. Consequently, intervention focused on this group could take the form of a preschool intellectual enrichment program. These prenatal and postnatal interventions could be carried out on an unselected group of mothers and their offspring. Alternatively, they could be targeted on a group of mothers who are deemed to be "at risk" for pregnancy and birth complications or who are thought to be more likely to have violent offspring on the basis of demographic and social characteristics or on the basis of crime and violence in the child's father. Another possibility is that experimental groups of both high-risk and low-risk mothers
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 384 could be included to assess the effects of possible interactions between prenatal and perinatal complications and social variables. Since socioeconomic disadvantage increases the risk of school failure in low-birthweight children, it might be expected that interventions would be most effective in socially disadvantaged families. Although it would take many years to evaluate the ultimate effect of such a health care program on violent adult behavior, initial assessments of success could be made relatively early in life. For example, the effects of this intervention on early temperamental antecedents of aggression could be assessed at age 20-30 months. Ratings of aggressive play behavior could be made at 3-4 years of age, while aggressive behavior in the home and school could be assessed by parents, teachers, and peers in early childhood. By adolescence, it would also be possible to assess self-reported violent behavior and to collect school data and public records. An important advantage of such an intervention is that it would be expected to result in a wide range of health benefits to the individual in addition to any effects on criminal behavior and violence. Social Learning and Cognitive-Behavioral Interventions Many studies have been carried out using variants of cognitive-behavioral therapy and interpersonal social skills training with aggressive children, as well as parent training and family therapy (for examples, see the reviews in Crowell et al., 1987; Keith, 1984). Behavioral management interventions based on social learning theory aim to alter the contingencies between responses and reinforcement, since aggression is viewed as inadvertently reinforced or modeled. Several interventions based on this theory are discussed below. Television Violence Since research indicates that television violence viewing in childhood predicts adult violence (Eron, 1987), interventions have been conducted to mitigate the possible effects of television violence on aggression. These either attempt to change the child's attitudes about television violence and the appropriateness of aggressive behavior (e.g., Huesmann et al., 1983) or use parent training to help control the child's viewing habits (e.g., Singer and Singer, 1981). Studies have attempted to increase prosocial behavior in children
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 385 by encouraging children to watch television programs that emphasize the prosocial behavior of the characters (Eron, 1986; Eron and Huesmann, 1984). Although these interventions have received some success in the short term, no long-term effects have been demonstrated. Furthermore, the observed effects have been confined to the laboratory and have focused almost exclusively on children from preschool ages to age 10 (Eron, 1986). Some intervention strategies appear to be more successful than others. For example, interventions that work directly with children (e.g., Huesmann et al., 1983) appear to be more successful than interventions that approach the child's behavior indirectly (e.g., Singer and Singer's attempt to alter the child's behavior by intervening with the parent). Future interventions in this area might therefore usefully focus on changing the child's attitudes about television violence and the appropriateness of aggressive behavior using direct methods. Cognitive/Behavioral Interventions A major difficulty facing all intervention programs is that aggressive behavioral strategies may be well learned early in the child's life and in consequence may be relatively intractable. Nevertheless, programs that have combined cognitive and behavioral approaches have proved to be the most effective forms of intervention and may be particularly effective with those children at risk of developing aggressive and violent behavior (Eron, 1986). However, Eron has argued that, for such programs to work, they must both emphasize the undesirability of aggressive behavior and also provide the child with alternative problem-solving behaviors. The success of interventions based on social-cognitive theory may be a function of the context in which the program is administered. For example, Guerra (1990) reported different success for two interventions. The first program aimed to develop social problem-solving skills, change beliefs about aggression, change attitudes toward television, and foster prosocial behavior in black and Hispanic aggressive schoolchildren with low socioeconomic status in the second to fourth grades. It was based on the social information processing theory of Dodge (1986). The intervention program produced no short-term behavioral change in this sample. However, an intervention also based on a social-cognitive training program for seriously aggressive, incarcerated 15- to 18-year-old delinquents resulted in posttreatment reductions in antisocial behavior.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 386 The facts that the latter population was older and that the study was conducted in an environment in which there were major incentives to reduce antisocial behavior represent contextual factors that may have led to a more successful outcome (Guerra, 1990). It may be of some value therefore for future intervention programs to pay attention to the role such contextual factors might play in mediating the success or failure of the intervention. It is common for many cognitive-behavioral intervention projects to show a short-term reductive effect on childhood aggression. However, there are few such studies with large samples, an experimental design, and a long-term follow-up to investigate possible effects on adult violence. Furthermore, most interventions have been carried out on nonclinical samples; research is needed to assess more fully whether success generalizes to more severely disturbed populations of aggressive children (Kendall et al., 1991). Social Skills Training A number of studies have demonstrated that social skills training can have effects in changing aggressive behavior in children. Specific aspects of social skills training that appear to be particularly effective in altering behavior include social relations training, prohibition of aggression, anger control, and cognitive- behavioral problem solving (Pepler et al., 1991). Furthermore, some evidence exists to support the notion that social skills training can be generalized to other settings, and that the benefits of such training can be maintained over time (Kazdin, 1988). One example of a social skills intervention that has demonstrated some moderate behavioral improvements extending over time is the Earlscourt Social Skills Group Program (Pepler et al., 1991). This program, based on social learning theory and cognitive-behavioral theory, is an experiential program designed to improve the self-control and social skills of aggressive children ages 6-12. Eight basic social skills are taught in group sessions conducted twice a week for a period of 12-15 weeks: "problem solving," "know your feelings," "listening," "following instructions," âjoining in," "using self-control," "responding to teasing," and "keeping out of fights." The program makes use of a wide number of behavioral techniques, including positive reinforcement, behavioral rehearsal, generalization, and training parents in child management skills. Social-cognitive techniques include enhancing problem-solving skills, increasing awareness of feelings, and promoting thinking aloud to assist self- monitoring. Children in
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 387 this social skills group showed significant improvements in externalizing behavior problems (including aggression) as rated by teachers relative to a control group on the Achenbach Child Behavior Checklist, and these improvements were maintained over a three-month follow-up period. Although cognitive social skills programs have been generally effective, most of these have shown improvements in cognitive abilities in a laboratory setting rather than in antisocial behavior, and the extent of change has often been small (Kazdin, 1988). Clear delineation of the target sample, use of stringent selection criteria, and identification of subtypes on whom different varieties of cognitive social skills interventions work best, are further areas that future intervention studies should address (Kazdin, 1988). Jones and Offord (1989) reported some initial evidence of effectiveness in their PALS (Participate and Learn Skills) community project. This skill development program was offered to an experimental group of all children ages 5 to 15 in a publicly supported housing complex in Ottawa, Ontario for a period of 32 months. A control housing complex had available only minimal recreational services provided by the city department. The PALS project sponsored 40 programs in 25 skill areas. Many of these programs were sports-based, but they also included guitar, ballet, baton, scouting, and other nonsports activities. During the 32 months of the intervention, police charges against the juveniles at the experimental site were less than one-quarter of those at the control site, and in addition there was a marked reduction in security violations in the experimental compared with the control housing complex. There were also improvements in school behavior and self-esteem in the experimental group, but these changes were not statistically significant. A cost-benefit analysis showed that the potential savings in terms of reduced vandalism, police time, and fire costs greatly exceeded the costs of the program. Two important questions concerning this program remain unanswered: (1) Do such skills interventions have any long-term effects on levels of antisocial behavior? (2) Do they have specific effects on violence in particular? If these questions can be answered affirmatively, consideration should be given to the inclusion of skills training in intervention programs for violent behavior in children. A recent study by Tremblay et al. (1991) combined social skills training with other interventions in 7-year-old boys selected for disruptiveness and low socioeconomic status. Disruptive boys were randomly allocated to an experimental group receiving interventions,
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 388 an "observation" group who received no intervention but who completed questionnaires and took part in observational studies, and a control group. Treatment for the boys consisted of social skills training (e.g., instructions on what to do when angry or when teased), training in fantasy play focused on prosocial alternatives to the expression of anger, and training on critical television watching. This treatment was combined with parent training consisting of monitoring the child's behavior, use of positive reinforcement, and family crisis management. Outcome measures (ratings by teachers, mothers, peers, and self- reported delinquency) were taken on three occasions following completion of the two-year program for a period of 27 months. No positive long-term treatment effects were observed for teacher, mother, or peer ratings. For self-reported delinquency however, treated boys reported reductions in fighting and stealing. Furthermore, fewer treated boys had to attend special educational classes or were held back in school. The fact that reductions in fighting were not immediately observed but did emerge at the longest follow- up period suggests that some intervention effects may take time to work through. Again, a replication and long-term follow-up of the effects of this multimodal intervention on adult violence are required in order fully to assess the benefits of this type of intervention. Parent Training Since poor parental childrearing techniques predict delinquency (Farrington and West, 1990), intervention programs that attempt to train parents may play a general role in reducing delinquency and may also result in some reduction in later violent offending. Patterson (1980; Paterson et al., 1982) viewed violence as in part a coercive act by the child to gain attention from the parent or as a mechanism to reduce frustration. In a parent intervention program aimed at families with delinquent and predelinquent children, parents were trained in effective childrearing methods, such as noticing what a child was doing, monitoring behavior over long periods, clearly stating house rules, making rewards and punishments contingent on behavior, and negotiating disagreements so that conflicts and crises did not escalate. Similarly, children were taught that coercive behavior would not be tolerated and would be met with time-out (isolation) procedures. Patterson's treatment program has been shown to be effective in reducing antisocial child behavior over short periods in small-scale studies, and
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 389 consequently such programs may warrant inclusion in any future intervention strategy for violence. Parent training programs have been evaluated in hundreds of outcome studies (Kazdin, 1985) and have been shown to be effective in reducing aggression and conduct disorder in children. As Kazdin (1988) points out, however, brief interventions tend to show fewer benefits, and families characterized by marital discord show fewer treatment gains, while children from families of low socioeconomic status are less likely to maintain these advantages over time. These limitations are not trivial, since aggressive children who are destined to become violent adults are more likely to come from socially disadvantaged backgrounds. Future parent training intervention programs must attempt to address these issues in addition to demonstrating the long-term effectiveness of these programs. Intervention programs that combine parent training with teacher training may be particularly beneficial because they tackle aggressive behavior in two crucial social settings. Hawkins et al. (1991) conducted such a program based on a social developmental model integrating social control theory with social learning theory (Hawkins and Weis, 1985). First-grade children in eight public elementary schools were subjects, with teacher ratings of antisocial and aggressive behaviors (assessed using the Achenbach Child Behavior Checklist) used as short-term outcome measures taken at the end of the second grade. Parent training consisted of the use of appropriate rewards and punishments, consistent discipline, use of effective communication skills, and skills to involve children in family activities. Teacher training consisted of proactive classroom management methods, cognitive social skills training, and interactive teaching methods. Boys who received exposure to these interventions were found to have lower scores on aggressive behavior and externalizing behavior problems than controls. Girls in the experimental group showed reductions in self-destructive behavior and depression, but no differences in aggression were observed. Importantly, the effectiveness of this dual intervention in reducing aggression in white males was not replicated in black males. Two important questions are raised by this study. The first is whether combining parent and teacher intervention programs will be more successful than single interventions in the long term. The second is whether the types of intervention programs developed for white children are appropriate for black children; it may be that special programs need to be developed to suit the affective
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 390 and cognitive styles of black children. Future intervention studies in this area must clearly pay close attention to the issue of race differences both in levels of aggressive behavior and in intervention effectiveness. School-Based Interventions Intellectual Enrichment Farrington and West (1990) have argued that school failure is an important predictor of later offending, and that preschool intellectual enrichment programs show some evidence of reductions in both school failure and later offending. For example, Berrueta-Clement et al. (1984) reported that children who received a preschool intellectual enrichment program were less likely to be arrested or charged, more likely to graduate from high school, and more likely to be employed at age 19 than controls. Future school-based interventions could usefully incorporate preschool intellectual enrichment programs, particularly targeted on those who are at risk for later crime and violence. U.S. Public Law 99-457 mandates the states to provide similar programs for at-risk children ages 3-5 (Short et al., 1990). Because school failure is a consistent predictor (or correlate) of violent behavior, it may be useful to target children who are at risk of school failure by virtue of living in high-risk neighborhoods. One possible intervention would be to select neighborhoods with high rates of both school failure and delinquency and to test every 4-year-old boy with a battery that assesses knowledge of the alphabet and of numbers and short-term memory. The administration of such a battery would require 45 minutes. About 15 percent of the children tested might show minimal knowledge of the alphabet and numbers and a poor short-term memory; these children are at risk for early school failure. The intervention would be run during the kindergarten year, perhaps by high school seniors who see the children once a day, five days a week, for the academic year. Some communities (Oakland, California, and San Antonio, Texas) have reduced the rate of school failure in this way. However, they did not assess the effects on conduct disorder, although it is quite conceivable that this intervention might lead to a reduction in antisocial behavior and later violence.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 391 Social Relations The notion that social rejection may predispose children to aggression has led to school-based interventions aimed at increasing the quality of children's social relations with their peers. Coie et al. (1991) report a school-based intervention made up of social problem solving, positive play training, group entry skill training, and dealing effectively with strong negative feelings. Although aggressive-rejected children showed some improvements, few statistically significant differences were observed. Nevertheless, consideration could be given to developing school-based violence prevention programs, aimed specifically at children in elementary and junior high school grades, together with community-based child development centers that provide family support and educational preparation for preschool children. Bullying A national intervention campaign against bullying was initiated in Norway in 1983. The program emphasized positive involvement from teachers and parents, setting firm limits on unacceptable behavior, and the use of nonhostile, noncorporal sanctions on rule violations. Not only did bullying decrease by 50 percent over the course of two years, but thefts, vandalism, and truancy rates were also reduced, while student satisfaction with school life increased. Effects were the same for both boys and girls, and aggressive bullying behavior was not found to be displaced to other situations outside the confines of the school (Olweus, 1991). An important question concerns whether the success of this type of intervention in Scandinavia would generalize to aggressive behavior in U.S. schools. This program is currently being tested in Canada, England, and the Netherlands. Strategies and Issues in the Implementation of Intervention Programs Research on the prevention of aggressive and violent behavior needs to acknowledge the multiple causes and systems involved in the development of such behavior. Previous interventions based on single variables or sets of variables have met with limited success, and a more successful strategy for preventing later violence might involve simultaneously targeting the individual and other areas such as school, peers, and family factors that are amenable to change (Guerra, 1990). The different elements of a global
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 392 intervention package might interact to produce a disproportionate effect. An important issue in intervention research is what individual form of treatment is best able to alter aggressive behavior in children. One way of addressing this is to conduct studies that compare several forms of experimental interventions. It has been argued that single modality treatments are less effective in reducing antisocial behavior than multimodal treatment approaches (Kazdin, 1988). It is desirable to establish not only the effectiveness of a multimodal intervention package but also which individual aspects of the package are responsible for behavioral improvement. Hence, it is recommended that wherever possible multimodal treatment evaluations be complemented by analyses of which components of the package are most effective, and that the types of treatments that are packaged together are carefully chosen and conceptually justified with respect to the target problem at hand. Many intervention researchers have failed to base their intervention strategies on clearly developed theoretical models of aggression; consequently, the outcomes of these studies have not been encouraging (Coie et al., 1991). It is recommended that future interventions be developed from a sound theoretical basis in order that intervention studies can inform theories of violent behavior. Farrington and West (1990) have found that the worst offenders in their longitudinal study tended to come from the poorest families with the worst housing. For example, low family income measured at age 8-10 was found to be the best predictor of general social failure at age 32. These findings suggest that one useful strategy may be to target economic resources to the very poorest families in interventions aimed at reducing levels of crime and violence. The notion that childhood aggression may be differentiated into proactive and reactive types (Dodge, 1991) has implications for intervention. Dodge (1991) argued that most intervention approaches are implemented without regard for the type of aggressive behavior under scrutiny, and that different types of aggression are likely to respond differently to different types of intervention. Reactive aggressives who overattribute hostility to others in provocative situations may respond best to treatment aimed at training them to understand better others' thoughts and feelings. Proactively aggressive children may respond more favorably to consistent punishment of aggressive behavior and reinforcement of nonaggressive responses; this latter group may also have a better prognosis than
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 393 the former group. Three intervention programs cited by Dodge (1991) as being particularly suitable for differential implementation with these two types of aggressive children include social problem-solving skills training (Spivack and Shure, 1974), angercontrol training (Lochman and Curry, 1986), and parent training (Patterson et al., 1982). In a similar fashion, Kendall et al. (1991) have argued that conduct disordered children with lower levels of hostility and aggression are especially likely to benefit from cognitive-behavioral therapy. Research on sequences and on prediction might help in identifying when and how it might be best to intervene to prevent the development of adult violence. For example, interventions might be targeted just before key developmental transitions (e.g., from less serious aggression to more serious violence) or when the correlation between the predictor and the outcome is still relatively low (indicating malleability). A related issue concerns at what age interventions may be most successfully imposed. There is some evidence that interventions conducted early in the developmental process are more effective than later interventions (Hawkins et al., 1988), although some exceptions have been noted (Guerra, 1990). From a cost-benefit perspective, it may be better to intervene at a later developmental stage, when one can be more sure that the targeted group represents those with a high likelihood of going on to become violent adults. Some data by Farrington and Hawkins (1991) indicate that childhood events are more important than teenage behavior in predicting the persistence of offending in the twenties, and they speculate that early prevention may have a greater potential than later criminal justice measures for reducing adult crime. Prediction studies that can push back the age at which adult violence can be predicted are clearly important, in that they will help determine at what developmental stage interventions may be most profitably conducted. Ideas about intervention possibilities can also be obtained by studying aggressive children who do not become violent adults, and seeking to identify factors that protect them from making the progression. Similarly, most violent offenders desist from violent offending in their twenties, but there has been little research investigating the factors that foster such desistance. Some of it may involve switching to other kinds of activities. For example, aggressive juveniles tend to be unemployed and to be heavy drinkers in their twenties, even though their aggressive behavior has declined (Farrington, 1991b).
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 394 Attention should also be paid to intervening with those children who possess multiple predisposing factors for later aggression and violence. For example, one intervention that may be of value consists of studying children ages 4-5 with uninhibited (fearless) temperaments who are also at risk of early school failure. The combination of a young child who is minimally anxious and is about to fail a major life task puts that child at high risk for the development of antisocial behavior. Intervention with such children before school entrance could be of great value. Policy Issues The results of intervention research have important implications for public policy. The likely effects of large-scale policy changes should be evaluated beforehand in small-scale experiments. Some policy questions that are likely to arise in connection with the development of an individual potential for violence are as follows: (1) Should aggressive children be identified at an early age (under 10) and given ameliorative intervention programs such as focusing on skills training, anger control (Novaco), and parent training (Patterson et al., 1991)? (2) Should pregnant women who are at risk for pregnancy and birth complications be provided with better antenatal services in order to reduce the probability of violence by their children in later life? (3) Should more widespread programs be introduced, targeted on all children rather than those identified as aggressive? For example, the Olweus antibullying campaign is a wide-ranging program, and so is the attempt to convey prosocial values through prosocial models (e.g., Mr. Rogers). References American Psychiatric Association 1987 Diagnostic and Statistical Manual of Mental Disorders (3rd ed., revised). Washington, D.C.: American Psychiatric Association. Baker, L.A. 1986 Estimating genetic correlations among discordant phenotypes: An analysis of criminal convictions and psychiatric hospital diagnoses in Danish adoptees. Behavior Genetics 16:127-142.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 395 Bandura, A. 1973 Aggression: A Social Learning Analysis. Englewood Cliffs, N.J.: Prentice-Hall. Berkowitz, L. 1978 Is criminal violence normative behavior? Journal of Research in Crime and Delinquency 15:148-161. 1983 Aversively stimulated aggression: Some parallels and differences in research with animals and humans. American Psychologist 38:1135-1144. Berrueta-Clement, J.R., L.J. Schweinhart, W.S. Barnett, A.S. Epstein, and D.P. Weikart 1984 Changed Lives. Ypsilanti, Mich.: High/Scope. Blumstein, A., J. Cohen, J.A. Roth, and C.A. Visher, eds. 1986 Criminal Careers and "Career Criminals." Washington, D.C.: National Academy Press. Clarke, R.V., and D.B. Cornish 1985 Modelling offenders' decisions. Pp. 147-185 in M. Tonry and N. Morris, eds., Crime and Justice. Vol. 6. Chicago: University of Chicago Press. Coie, J.D., M. Underwood, and J.E. Lochman 1991 Programmatic intervention with aggressive children in the school setting. Pp. 389-410 in D.J. Pepler and K.H. Rubin, eds., The Development and Treatment of Childhood Aggression. Hillsdale, N.J.: Erlbaum. Comstock, G., and H. Paik 1990 The Effects of Television Violence on Aggressive Behavior: A Meta-Analysis. Unpublished report to the National Academy of Sciences Panel on the Understanding and Control of Violent Behavior , Washington, DC. Crowell, D.H., I.M. Evans, and C.R. O'Donnell, eds. 1987 Childhood Aggression and Violence. New York: Plenum. DiLalla, L.F., and I.I. Gottesman 1991 Biological and genetic contributors to violenceâWidom's untold tale. Psychological Bulletin 109:125-129. Dodge, K.A. 1980 Social cognition and children's aggressive behavior. Child Development 51:162-170. 1986 A social information processing model of social competence in children. Pp. 77-125 in M. Perlmutter, ed., Minnesota Symposium in Child Psychology. Vol. 18. Hillsdale, N.J.: Erlbaum. 1991 The structure and function of reactive and proactive aggression. Pp. 201-218 in D.J. Pepler and K.H. Rubin, eds., The Development and Treatment of Childhood Aggression. Hillsdale, N.J.: Erlbaum. Dodge, K.A., and J.D. Coie 1987 Social information processing factors in reactive and proactive
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 396 aggression in children's peer groups. Journal of Personality and Social Psychology 53:1146-1158. Dodge, K.A., and J.P. Newman 1981 Biased decision-making processes in aggressive boys. Journal of Abnormal Psychology 90:375-379. Dollard, J., L.W. Doob, N.E. Miller, O.H. Mowrer, and R.R. Sears 1939 Frustration and Aggression. New Haven: Yale University Press. Drillien, C.M. 1964 The Growth and Development of the Premature Infant. Baltimore: Williams and Wilkins. Elliott, D.S., D. Huizinga, and S. Menard 1989 Multiple Problem Youth. New York: Springer-Verlag. Eron, L.D. 1986 Interventions to mitigate the psychological effects of media violence on aggressive behavior. Journal of Social Issues 42:155-169. 1987 The development of aggressive behavior from the perspective of a developing behaviorism. American Psychologist 42:435-442. 1990 Understanding aggression. Presidential address to the World Meeting of ISRA, Banff, Alberta, June 12. Eron, L.D., and L.R. Huesmann 1984 The relation of prosocial behavior to the development of aggression and psychopathology. Aggressive Behavior 10:201-212. 1987 The stability of aggressive behavior in cross-national comparisons. Pp. 207-218 in C. Kagitcibasi, ed., Growth and Progress in Cross-Cultural Psychology. Lisse, Holland: Swets and Zeitlinger. 1990 The stability of aggressive behaviorâEven unto the third generation. Pp. 147-156 in M. Lewis and S.M. Miller, eds., Handbook of Developmental Psychopathology. New York: Plenum Press. Eron, L.D., L.R. Huesmann, P. Brice, P. Fischer, and R. Mermelstein 1983 Age trends in the development of aggression, sextyping, and related television habits. Developmental Psychology 19:71-77. Eron, L.D., L.R. Huesmann, and A. Zelli 1991 The role of parental variables in the learning of aggression. Pp. 169-188 in D. Pepler and K.H. Rubin, eds., The Development and Treatment of Childhood Aggression. Hillsdale, N.J.: Erlbaum. Farrington, D.P. 1987 Implications of biological findings for criminological research. Pp. 42-64 in S.A. Mednick, T.E. Moffitt, and S.E. Stack, eds., The Causes of Crime: New Biological Approaches. Cambridge: Cambridge University Press. 1988 Advancing knowledge about delinquency and crime: The need for a coordinated program of longitudinal research. Behavioral Sciences and the Law 6:307-331. 1989 Early predictors of adolescent aggression and adult violence. Violence and Victims 4:79-100.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 397 1990 Age, period, cohort, and offending. Pp. 51-75 in D.M. Gottfredson and R.V. Clarke, eds., Policy and Theory in Criminal Justice. Aldershot, England: Avebury. 1991a Antisocial personality from childhood to adulthood. The Psychologist 4. 1991b Childhood aggression and adult violence: Early precursors and later-life outcomes. Pp. 5-29 in D.J. Pepler and K.H. Rubin, eds., The Development and Treatment of Childhood Aggression. Hillsdale, N.J.: Erlbaum. Farrington, D.P., and J.D. Hawkins 1991 Predicting participation, early onset, and later persistence in officially recorded offending. Criminal Behavior and Mental Health 1:1-33. Farrington, D.P., and R. Loeber 1989 RIOC (Relative Improvement Over Chance) and phi as measures of predictive efficiency and strength of association in 2X2 tables. Journal of Quantitative Criminology 5:201-213. Farrington, D.P., and D.J. West 1990 The Cambridge study in delinquent development: A long-term follow-up study of 411 London males. Pp. 115-138 in H.J. Kerner and G. Kaiser, eds., Criminality: Personality, Behavior, and Life History. Berlin: Springer-Verlag. Farrington, D.P., B. Gallagher, L. Morley, R.J. St. Leger, and D.J. West 1988 Are there any successful men from criminogenic backgrounds? Psychiatry 51:116-130. Federal Bureau of Investigation 1982 Uniform Crime Reports. Washington, D.C.: U.S. Department of Justice. Fogel, C.A., S.A. Mednick, and N. Michelson 1985 Hyperactive behavior and minor physical anomalies. Acta Psychiatrica Scandinavia 72:551-556. Freedman, J.L. 1986 Television violence and aggression: A rejoinder. Psychological Bulletin 100:372-378. Friedrich-Cofer, L., and A.C. Huston 1986 Television violence and aggression: The debate continues. Psychological Bulletin 100:364-371. Guerra, N. 1990 Social cognitive approaches to the prevention of antisocial behavior in children. Paper presented at the Workshop on Hostility and Sociability, Warsaw, Poland (September). Hamparian, D.M., R. Schuster, S. Dinitz, and J.P. Conrad 1978 The Violent Few. Lexington, Mass.: Heath. Hamparian, D.M., J.M. Davis, J.M. Jacobson, and R.E. McGraw 1985 The Young Criminal Years of the Violent Few. Washington, D.C.: National Institute for Juvenile Justice and Delinquency Prevention.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 398 Hare, R.D. 1978 Electrodermal and cardiovascular correlates of psychopathy. Pp. 107-144 in R.D. Hare and D. Schalling, eds., Psychopathic Behavior: Approaches to Research. New York: Wiley. 1981 Psychopathy and violence. In J.R. Hays, K. Roberts, and K. Solway, eds., Violence and the Violent Individual. New York: Spectrum. Hare, R.D., L.M. McPherson, and A.E. Forth 1988 Male psychopaths and their criminal careers. Journal of Consulting and Clinical Psychology 56:710-714. Hawkins, J.D., and J.G. Weis 1985 The social development model: An integrated approach to delinquency prevention. Journal of Primary Prevention 6:73-97. Hawkins, J.D., J.M. Jenson, R.F. Catalano, and D.M. Lishner 1988 Delinquency and drug abuse: Implications for social services. Social Service Review 258-284. Hawkins, J.D., E. Von Cleve, and R.F. Catalano 1991 Reducing early childhood aggression. Journal of the American Academy of Child and Adolescent Psychiatry. Heath, L., L.B. Bresolin, and R.C. Rinaldi 1989 Effects of media violence on children. Archives of General Psychiatry 46:376-379. Huesmann, L.R., and L.D. Eron 1984 Cognitive processes and the persistence of aggressive behavior. Aggressive Behavior 10:243-251. 1986 Television and the Aggressive Child: A Cross-Cultural Comparison. Hillsdale, N.J.: Erlbaum. 1988 Early predictors of criminality. Paper presented at XXIV International Congress of Psychology, Sydney, Australia. 1989 Individual differences and the trait of aggression. European Journal of Personality 3:95-106. Huesmann, L.R., L.D. Eron, R. Klein, P. Brice, and P. Fischer 1983 Mitigating the imitation of aggressive behaviors by changing children's attitudes about media violence. Journal of Personality and Social Psychology 44:899-910. Huesmann, L.R., L.D. Eron, M.M. Lefkowitz, and L.O. Walder 1984 The stability of aggression over time and generations. Developmental Psychology 20:1120-1134. Jones, M.B., and D.R. Offord 1989 Reduction of antisocial behavior in poor children by non-school skill-development. Journal of Child Psychology and Psychiatry 30:737-750. Kagan, J. 1980 Perspectives on continuity. In O.G. Brim and J. Kagan, eds., Constancy and Change in Human Development. Cambridge, Mass.: Harvard University Press.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 399 1989 Temperamental contributions to social behavior. American Psychologist 44:668-674. 1991 The adolescent at risk. Paper presented at a conference âAdolescents at Risk," Cornell University Medical College, New York, February. Kandel, D.B., and R. Faust 1975 Sequence and stages in patterns of adolescent drug use. Archives of General Psychiatry 32:923-932. Kazdin, A.E. 1985 Treatment of Antisocial Behavior in Children and Adolescents. Homewood, Ill.: Dorsey Press. 1988 Treatment of antisocial behavior in children: Current status and future directions. Psychological Bulletin 102:187-203. Keith, C.R., ed. 1984 The Aggressive Adolescent. New York: Free Press. Kellam, S.G., C.H. Brown, B.R. Rubin, and M.E. Ensminger 1983 Paths leading to teenage psychiatric symptoms and substance use. In S.B. Guze, F.J. Earls, and J.E. Barratt, eds., Childhood Psychopathology and Development. New York: Raven Press. Kendall, P.C., K.R. Ronan, and J. Epps 1991 Aggression in children/adolescents: Cognitive- behavioral treatment perspectives. Pp. 341-360 in D.J. Pepler and K.H. Rubin, eds., The Development and Treatment of Childhood Aggression. Hillsdale, N.J.: Erlbaum. Lewis, D.P., S.S. Shanok, J.H. Pincus, and G.H. Glaser 1979 Violent juvenile delinquents. Journal of the American Academy of Child Psychiatry 18:307-319. Litt, S.M. 1971 Perinatal complications and criminality. Doctoral dissertation, University of Michigan. Lochman, J.E., and J.F. Curry 1982 Effects of social problem-solving training and self-instruction with aggressive boys. Journal of Clinical Child Psychology 15:159-164. Magnusson, D., H. Stattin, and V.L. Allen 1986 Differential maturation among girls and its relations to social adjustment: A longitudinal perspective. Pp. 135-172 in P.B. Baltes, D.L. Featherman, and R.M. Lerner, eds., Life-Span Development and Behavior. Hillsdale, N.J.: Erlbaum. Margolin, G., L. Gorin-Sibner, and L. Gleberman 1988 Wife battering. In V.B. Van Hasselt, R.L. Morrison, A.S. Bellack, and M. Hersen, eds., Handbook of Family Violence. New York: Plenum. McCord, J. 1980 Patterns of deviance. Pp. 157-167 in S.B. Sells, R. Crandall, M. Roff, J.S. Strauss, and W. Pollin, eds., Human Functioning in Longitudinal Perspective. Baltimore, Md.: Williams and Wilkins.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 400 1987 Aggression and shyness as predictors of problems. Paper presented at the Biennial Meeting of the Society for Research in Child Development, Baltimore. McCord, W., and J. McCord 1956 Psychopathy and Delinquency. New York: Grune and Stratton. McDermott, N.J., and M.J. Hindelang 1981 Juvenile Criminal Behavior in the United States. Washington, D.C.: U.S. Government Printing Office. Mednick, S.A., and E. Kandel 1988 Genetic and perinatal factors in violence. Pp. 121-134 in S.A. Mednick and T. Moffitt, eds., Biological Contributions to Crime Causation. Holland: Martinus Nijhoff. Megargee, E.I. 1973 Recent research on overcontrolled and undercontrolled personality patterns among violent offenders. Social Symposium 9:37-50. Miller, S.J., S.D. Dinitz, and J.P. Conrad 1982 Careers of the Violent. Lexington, Mass.: Lexington Books. Milner, J.S. 1986 The Child Abuse Potential Inventory: Manual. 2nd ed. Webster, N.C.: Psytec Corporation. 1990 Perpetrator and familial characteristics in physical and sexual child abuse cases. Department of Psychology, Northern Illinois University. Milner, J.S., K.R. Robertson, and D.L. Rogers 1990 Childhood history of abuse and child abuse potential. Journal of Family Violence 5:15-34. Moffitt, T.E. 1990 The neuropsychology of juvenile delinquency: A critical review. Pp. 99-169 in M. Tonry and N. Morris, eds., Crime and Justice, Vol. 12. Chicago: University of Chicago Press. Moffitt, T.E., S.A. Mednick, and W.F. Gabrielli 1989 Predicting careers of criminal violence: Descriptive data and dispositional factors. In D.A. Brizer and M. Crowner, eds., Current Approaches to the Prediction of Violence. Washington, D.C.: American Psychiatric Press. Mungas, D. 1983 An empirical analysis of specific syndromes of violent behavior. Journal of Nervous and Mental Diseases 171:354-361. Oakley, A., L. Rajan, and A. Grant 1990 Social support and pregnancy outcome. British Journal of Obstetrics and Gynaecology 97:155-162. Olweus, D. 1979 Stability of aggressive reaction patterns in males: A review. Psychological Bulletin 86:852-875. 1980 Familial and temperamental determinants of aggressive behavior
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 401 in adolescent boys: A causal analysis. Developmental Psychology 16:644-660. 1987 In S.A. Mednick, T.E. Moffitt, and S.A. Stack, eds., The Causes of Crime: New Biological Approaches. Cambridge, England: Cambridge University Press. 1991 Bully/victim problems among schoolchildren: Basic facts and effects of a school based intervention program. Pp. 411-448 in D.J. Pepler and K.H. Rubin, eds., The Development and Treatment of Childhood Aggression. Hillsdale, N.J.: Erlbaum. Patterson, G.R. 1980 Children who steal. Pp. 73-90 in T. Hirschi and M. Gottfredson, eds., Understanding Crime. Beverly Hills, Calif.: Sage Publications. 1986 Performance models for antisocial boys. American Psychologist 41:432-444. Patterson, G.R., P. Chamberlain, and J.B. Reid 1982 A comparative evaluation of a parent training program. Behavior Therapy 13:638-650. Patterson, G.R., D. Capaldi, and L. Bank 1991 An early starter model for predicting delinquency. Pp. 139-168 in D.J. Pepler and K.H. Rubin, eds., The Development and Treatment of Childhood Aggression. Hillsdale, N.J.: Erlbaum. Pepler, D.J., and K.H. Rubin, eds. 1991 The Development and Treatment of Childhood Aggression. Hillsdale, N.J.: Erlbaum. Pepler, D.J., G. King, and W. Byrd 1991 A social-cognitively based social skills training program for aggressive children. Pp. 361-379 in D.J. Pepler and K.H. Rubin, eds., The Development and Treatment of Childhood Aggression. Hillsdale, N.J.: Erlbaum. Raine, A., and F. Jones 1987 Attention, autonomic arousal, and personality in behaviorally disordered children. Journal of Abnormal Child Psychology 15:583-599. Raine, A., and P. Venables In Press Antisocial behavior: Evolution, genetics, neuropsychology, and psychophysiology. In A. Gale and M. Eysenck, eds., Handbook of Individual Differences. Chichester: Wiley. Reznick, J.S., J. Kagan, N. Snidman, M. Gerstein, K. Baak, and A. Rosenberg 1986 Inhibited and uninhibited children: A follow-up study. Child Development 57:660-680. Richard, B.A., and K.A. Dodge 1982 Social maladjustment and problem solving in school aged children. Journal of Consulting and Clinical Psychology 50:226-233.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 402 Robins, L.N. 1991 Antisocial personality. Pp. 259-290 in L.N. Robins and D. Regier, eds., Psychiatric Disorder in America. New York: Macmillan/Free Press. Rutter, M., and H. Giller 1983 Juvenile Delinquency: Trends and Perspectives. Harmondsworth: Penguin. Shannon, L.W. 1988 Criminal Career Continuity. New York: Human Sciences Press. Short, R.J., R.J. Simeonsson, and G.S. Huntington 1990 Early intervention: Implications of Public Law 99-457 for professional child psychology. Professional Psychology 21:88-93. Singer, J.L., and D.G. Singer 1981 Television, Imagination, and Aggression: A Study of Pre- Schoolers. Hillsdale, N.J.: Erlbaum. Solway, K.S., L. Richardson, J.R. Hays, and V.H. Elcon 1981 Adolescent murderers. Pp. 193-209 in J.R. Hays, T.K. Roberts, and K.S. Solway, eds., Violence and the Violent Individual. Jamaica, N.Y.: Spectrum. Spivack, G., and M.B. Shure 1974 Social Adjustment of Young Children: A Cognitive Approach to Solving Real-Life Problems. San Francisco, Calif.: Jossey-Bass. Stark, E., and A. Flitcraft 1982 Medical therapy as repression: The case of the battered woman. Health and Medicine 1:29-32. Stattin, H., D. Magnusson, and H. Reichel 1989 Criminal activity at different ages: A study based on a Swedish longitudinal research population. British Journal of Criminology 29:368-385. Tonry, M., L.E. Ohlin, and D.P. Farrington 1991 Human Development and Criminal Behavior. New York: Springer-Verlag. Tremblay, R.E., J. McCord, H. Boileau, P. Charlebois, C. Cagnon, M. LeBlanc, and S. Larivee 1991 Can disruptive boys be helped to become competent? Psychiatry. U.S. Department of Health, Education, and Welfare 1978 Violent Schools - Safe Schools. Washington, D.C.: National Institute of Education. Virkkunen, M. 1987 Metabolic dysfunctions among habitually violent offenders: Reactive hypoglycemia and cholesterol levels. In S.A. Mednick, T.E. Moffitt, and S. Stack, eds., The Causes of Crime: New Biological Approaches. New York: Cambridge.
THE DEVELOPMENT OF AN INDIVIDUAL POTENTIAL FOR VIOLENCE 403 Wadsworth, M.E.J. 1976 Delinquency, pulse rate and early emotional deprivation. British Journal of Criminology 16:245-256. Weiner, N.A. 1989 Violent criminal careers and "violent career criminals": An overview of the research literature. Pp. 35-138 in N.A. Weiner and M.E. Wolfgang, eds., Violent Crime, Violent Criminals. Newbury Park, Calif.: Sage Publications. Werner, E.E., and R.S. Smith 1982 Vulnerable but Invincible. New York: McGraw-Hill. Widom, C.S. 1989a Does violence beget violence? A critical examination of the literature. Psychological Bulletin 106:3-28. 1989b The cycle of violence. Science 244:160-166. 1991 A tail on an untold tale: Response to "Biological and genetic contributors to violenceâWidom's untold tale." Psychological Bulletin 109:130-132. Wikstrom, P.O. 1985 Everyday Violence in Contemporary Sweden. Stockholm: National Council for Crime Prevention. Wolfgang, M.E., R.M. Figlio, and T. Sellin 1972 Delinquency in a Birth Cohort. Chicago: University of Chicago Press. Wolfgang, M.E., T.P. Thornberry, and R.M. Figlio 1987 From Boy to Man, From Delinquency to Crime. Chicago: University of Chicago Press. Wood, W., F.Y. Wong, and J.G. Chachere 1991 Effects of media violence on viewers' aggression in unconstrained social interaction. Psychological Bulletin 109:371-383.